A Tool for Student Grade Prediction, Revision Direction Assistance, and School Support Recommendation with the Assistance of Neural Networks.
Abstract:
Student mental health is a problem that has been getting worse year on year in the UK. A part of this mental health problem is the huge amount of stress that is loaded upon students year in year out, much of which comes from exams and studying for them. To help one’s stress and to improve one’s grades, it is recommended to find a balanced studying habit that works for oneself, however this can be quite hard when suffering from mental health issues, leading to lowered grades, and spiralling mood. By using the concept of predicting student grades with AI, this paper proposes a pair of tools that students and teachers can use to predict their final grades based on their prior performance which also recommends to the students and teachers’ ways for them to improve their final grade by the largest possible margin achievable by them. A pair of neural network models were created and implemented into user-facing python programs, which allow the users to discover their predicted grades, and points out ways that the school could help the student or which previous test they should revise in order to gain the most marks for their future final exam. The paper found that the models created are very accurate in their predictions and that current university students are interested in the capabilities of the program. With more detailed studies and an expanded training dataset, this tool could become a powerful part of the teaching toolkit in the UK, allowing students to improve their grades and take away the stress of figuring out what they should be revising.
Download Dylan's Thesis